Placenta growth factor (PlGF) is a mitogen for endothelial cells that can potentiate the growth and permeabilizing effects on endothelium of vascular endothelial growth factor. Here we report that hypoxia induces the expression of both PlGF mRNA and protein in immortalized/transformed mouse embryonic fibroblasts (mEFs) and in NIH 3T3 cells. Importantly, the magnitude of the induction of PlGF expression by hypoxia is enhanced by the presence of oncogenic Ras. To investigate the transcriptional component of hypoxia-inducible PlGF expression, we cloned and sequenced a 1350-bp fragment of the 5′-flanking region of the mouse gene. Analysis of the promoter region indicated the presence of putative consensus sequences for known hypoxia-responsive regulatory sites, including metal response elements and Sp1-like sites. In the present study, we show that the induction of PlGF expression by hypoxia is dependent on the presence of the metal response element-binding transcription factor 1 (MTF-1). Thus, in mEFs with targeted deletions of both MTF-1 alleles, hypoxia-induced increases of PlGF mRNA and protein levels were greatly attenuated compared with those in wild-type mEFs. Moreover, transient transfection of a PlGF promoter reporter gene into NIH 3T3 cells resulted in hypoxia-responsive transcriptional activation of the reporter. Finally, ectopic expression of MTF-1 resulted in increased basal transcriptional activity of a PlGF promoter reporter. Together, these findings demonstrate that the PlGF gene is responsive to hypoxia and that this response is mediated by MTF-1. It remains to be determined whether this activation is the result of direct and/or indirect transcriptional activation by MTF-1. The stimulatory effect of oncogenic Ras on the induction of PlGF expression in hypoxic cells suggests that PlGF could be an important proangiogenic factor in the tumor microenvironment.
PlGF4 is a member of the VEGF family of proangiogenic factors (1). Both PlGF and VEGF form glycosylated homodimers that share significant homology at the amino acid level within their platelet-derived growth factor-like regions, including the conservation of eight cysteine residues involved in intra- and interchain disulfide bond formation (1, 2). hPlGF is also similar to VEGF (and platelet-derived growth factor) in that alternative splicing of the mRNA from a single copy gene produces different isoforms of the protein [PlGF-1, PlGF-2, and PlGF-3 (3, 4)]. PlGF-2, or PlGF170, differs from the other two hPlGF isoforms in that it contains an additional 21 basic amino acid region at the COOH terminus that confers a heparin-binding ability similar to that of VEGF165 (3, 5). However, whereas all VEGF isoforms bind to the tyrosine kinase receptors flt-1 and KDR/flk-1, PlGF homodimers are believed to bind only to flt-1 (see Ref. 6), suggesting a role for the homodimers in endothelial cell-cell or cell-matrix interactions (7). PlGF may also potentiate the angiogenic effects of VEGF (8) because the two proteins have the ability to form heterodimers with each other (2, 9, 10, 11) that can activate the KDR/flk-1 receptor (9). Like VEGF, PlGF-2 has also been shown to bind to and activate neuropilin-1 (6, 12), a receptor found on many cell types in addition to endothelial cells. This receptor can bind at least five different ligands including VEGF165 and has established roles in a wide variety of cellular processes including axonal guidance (13). These findings suggest that PlGF is a component of extensive combinatorial interactions involving the VEGF superfamily.
In contrast to the widespread distribution of VEGF, PlGF expression was originally considered to be restricted to cell types of placental origin such as primary cytotrophoblasts and in vitro differentiated syncytiotrophoblasts (e.g., see Ref. 14). However, accumulating evidence indicates that PlGF is synthesized in many nonplacental cells and other tissues including human thyroid, brain, lung, and skeletal muscle (reviewed in Ref. 15). PlGF is also highly expressed in dermal microvascular endothelial cells and retinal pericytes (16), whereas conditioned medium from cultured human keratinocytes was found to contain both PlGF homodimers and PlGF/VEGF heterodimers (11). It has also been reported that migrating keratinocytes from healing wounds contain significant levels of PlGF from the third to the seventh day after injury (11). Other studies indicate that PlGF may be involved in the inflammatory process because PlGF homodimers and PlGF/VEGF heterodimers were detected in the synovial fluid of patients with inflammatory arthopathies (17). PlGF is also expressed in several types of solid tumors (18, 19, 20, 21). Furthermore, PlGF is believed to be involved in the pathogenesis of PDR and other ischemic retinal diseases (22, 23, 24).
Oxygen deprivation (hypoxia) appears to be a common feature of many of the pathophysiological conditions in which PLGF is expressed. For example, hypoxia is a characteristic common to many solid tumors and is believed to contribute to increased expression of proangiogenic proteins (e.g., VEGF) and subsequent activation of angiogenesis (25, 26). However, the role of hypoxia in the regulation of PlGF expression is currently unclear because the majority of cell lines used to examine its expression appear to be unresponsive to oxygen deprivation (10, 16, 27, 28). The findings presented here demonstrate that immortalized/transformed fibroblasts express PlGF and that exposure of these cells to physiologically relevant levels of hypoxia results in significant inductions of the steady-state levels of both PlGF mRNA and protein. We also show that this activation is dependent on the presence of the redox-sensitive transcription factor MTF-1 (29, 30) and that the hypoxia-inducible increase in PlGF expression is due at least in part to increased transcription of the PlGF gene, particularly in the presence of oncogenic ras.
MATERIALS AND METHODS
Cell Culture and Hypoxia.
All cell cultures, including mEFs and NIH 3T3, JAR (a human choriocarcinoma cell line), and HepG2 (a human hepatoma cell line) cells, were maintained in DMEM containing 10% dialyzed fetal bovine serum. Both SV40 large TAg and TAg/Ras-transformed wild-type (MTF-1+/+) and MTF-1 null (MTF-1−/−) mEFs were used in these studies. To obtain the mEFs, primary embryonic fibroblasts were isolated from 12.5-day-old mouse embryos using standard techniques (31). PCR genotyping for MTF-1 null cells was performed using genomic DNA prepared from yolk sacs and early passages of the primary cells. Wild-type and MTF-1 null primary cells were then transfected with either 10 μg of plasmid CMV-TAg or 10 μg of plasmid CMV-TAg and 10 μg of plasmid c-H-ras(A) (32) per 100-mm-diameter cell culture plate. The plasmid CMV-TAg directs expression of TAg, and c-H-ras(A) directs expression of oncogenic human H-ras. Cell foci were isolated, and immortalized cell lines were derived. All cell lines were genotyped again by PCR analysis for the presence of the MTF-1 wild-type and null alleles, as well as for genomic integration of TAg and oncogenic H-ras.
For immunoblotting analyses, medium was removed, cells were washed with PBS, and then fresh DMEM without fetal bovine serum was added immediately before hypoxia treatment. Our methods for exposing cells to hypoxia (pO2 ≤ 0.1% relative to air at pO2 of approximately 21%) have been described in detail elsewhere (33). Other levels of hypoxia were achieved by continuous flow-though of specific levels of oxygen and 5% CO2 at 37°C, using a glass vacuum desiccator (34). After exposure to hypoxia, an analysis for cell viability using Alamar Blue indicated that the cells were not adversely affected by these treatments.
RT-PCR and Northern Analyses.
Two DNA primers for RT-PCR were synthesized based on the mPlGF cDNA sequence (GenBank accession number X80171; Ref. 35). The sequences of the sense and antisense primers were, respectively, 5′-TTTCTCAGGATGTGCTCTGTGAA-3′ and 5′-CCTGGTTACCTCCGGGAAATGAC-3′. These primers were designed to amplify a sequence spanning exons 4–7 (cDNA bases 473–619) of the PlGF gene, producing a 147-bp cDNA fragment by RT-PCR amplification of mPlGF mRNA. This sequence was chosen to detect any mouse homologues of the known human splice variants of PlGF mRNA, if such homologues were present in the mEF cells. One control without RNA and four RNA samples containing 250 ng of total RNA each from aerobic and hypoxic wild-type mEFs (4 and 8 h of hypoxia) were used for RT-PCR analysis. A GeneAmp recombinant Thermus thermophilus Reverse Transcriptase kit (Perkin-Elmer Applied Biosystems, Foster City, CA) was used according to the manufacturer’s protocol. Thirty-five cycles each of 10 s at 95°C and 15 s at 60°C were performed with a Perkin-Elmer GeneAmp System 2400, followed by a final incubation at 60°C for 7 min. The reaction products, along with DNA molecular weight markers of 517 to 75 bp, were resolved by nondenaturing 5% PAGE in Tris-borate EDTA buffer. The gel was stained with ethidium bromide, and bands were visualized on an UV transilluminator.
Our procedure for Northern analysis is described elsewhere (36); here, however, total RNA was isolated from cells by using a RNasy kit (Qiagen Corp., Valencia, CA). Blots were probed with either an EcoRI/NotI (1.1 kb) fragment of the mPlGF cDNA (Genome Systems, Mountain View, CA) or a PCR-generated PlGF cDNA fragment. To label the PlGF probe, we followed the amplification reaction protocol described above, but with only the antisense primer and [32P]dCTP and the other unlabeled deoxynucleotide triphosphates. Signals were quantified by video densitometry using a Lynx 4000 image analyzer (Applied Imaging, Sunnyvale, CA).
After treatment with hypoxia, conditioned media from the mEF cultures were concentrated approximately 14-fold by using Centraprep ultrafiltration units (Millipore Corp., Bedford, MA). Equivalent volumes (20 μl) of each concentrated sample in standard SDS loading buffer were analyzed by 12% SDS-PAGE. Blots were probed with a primary goat anti-mPlGF-2 antibody (R&D Systems, Minneapolis, MN), followed by an antigoat IgG horseradish peroxidase-conjugated secondary antibody (Dako, Carpenteria, CA). PlGF protein was detected using an ECL-PLUS Enhanced Chemiluminescence detection system according to the manufacturer’s protocol (Amersham Life Science, Arlington Heights, IL).
Inverse PCR of the PlGF Gene Promoter Region.
Our initial attempt to clone the promoter region of mPlGF used a Mouse GenomeWalker Kit (Clontech, Palo Alto, CA). This kit supplies mouse genomic DNA that has been cleaved with a variety of restriction enzymes and ligated with linkers containing primer sites for PCR amplification. Two nested PCR primers complementary to the 5′-end of the longest known mouse cDNA for PlGF (GenBank accession number X96793) were used. The two primers were complementary to bases 268–291 and 295–318 of this cDNA sequence. Only one 500-bp fragment was obtained using this kit. Because only slightly more than 200 bp of this sequence were from the upstream region of the known mPlGF cDNA, it was considered unlikely that this fragment would contain all of the necessary regulatory elements to function as a promoter in reporter gene studies. Therefore, inverse PCR was used to clone a larger sequence of the mPlGF gene. Mouse genomic DNA (1 μg) was cleaved with BamHI, EcoRI, HindIII, KpnI, SstI, or XbaI. Each of the cleaved DNA preparations was dissolved in 0.5 ml of 50 mm Tris-HCl (pH 7.8), 10 mm MgCl2, 20 mm DTT, and 1 mm ATP, and then 2 units of T4 DNA ligase (Life Technologies, Inc., Rockville, MD) were added. The ligation reactions were then incubated at 5°C for 16 h. This low DNA concentration and low temperature ligation procedure favors the formation of circularized products. Each ligated DNA sample was precipitated with 3 volumes of ethanol, dried, and dissolved in 10 μl of water.
The longest known sequence of the mouse cDNA for PlGF described above was used to design the following two PCR primers: (a) 5′-GAAGTCTGAAGATGCGGTTTCCTTC-3′; and (b) 5′-CACTCCTCACCAGTTTCCTCAGCCA-3′. The first primer is complementary to positions 5–29 of the PlGF cDNA sequence, and the second primer is the same as the sequence at positions 101–125. Because these primers would amplify DNA in opposite directions, they are able to produce a fragment of the upstream PlGF promoter from the circularized mouse genomic DNA. One-tenth (100 ng) of each of the five preparations of circularized mouse genomic DNA was used as the template for a PCR reaction using the Advantage 2 PCR Enzyme System (Clontech). In addition to the DNA template, each PCR reaction contained 100 μl of Advantage 2 PCR buffer, 0.2 mm each deoxynucleotide triphosphate, 1 μl of Advantage 2 polymerase mix, and 1 μm each of the two primers mentioned above. The reactions were amplified with a Perkin-Elmer GeneAmp PCR System 2400 using 35 PCR cycles (30 s at 94°C, 180 s at 72°C). After gel electrophoresis, it was determined that only the HindIII- and SstI-cut circularized genomic DNA fragments produced PCR products: a 3.0- and a 1.5-kb product, respectively. Because the known mouse cDNA sequence for PlGF has an SstI site in the 5′-untranslated region, most of this SstI inverse PCR fragment contained sequences upstream of the known cDNA. The 1.5-kb PCR product from the SstI-cut circularized genomic DNA was cloned into the Srf I site of pPCR-Script Amp SK(+) (Stratagene, San Diego, CA), and this template was sequenced from both strands using a PE Applied Biosystems Prism 310 Genetic Analyzer. Two separate clones were used for the sequencing, and they proved to be identical. The 3′-end of this sequence matched the 200-bp of promoter sequence found by using the Mouse GenomeWalker Kit described above. The orientation of the pPCR-Script clones allowed the use of the SstI site on the vector that was 3′ to the insert and the SstI site on the 5′-end of the promoter fragment to cut out a 1.5-kb fragment. This SstI fragment was next cloned into the SstI site of the luciferase-containing pGL3-Basic reporter vector (Promega, Madison, WI), and the correct orientation was verified by restriction analysis. This construct was designated as pmPlGF(1.5kb)-Luc. The promoter fragment in this clone corresponds to positions 1–1386 of the mPlGF gene (Fig. 4 A).
Cell Transfections and Treatments.
NIH 3T3 cells were plated at 5 × 104 cells/35-mm-diameter culture dish and grown to approximately 60–70% confluence before transfection using the Effectene Transfection Reagent (Qiagen) according to the manufacturer’s protocols. To determine the effect of hypoxia on PlGF transcription, the cells were transfected with 0.4 μg of pmPlGF(1.5kb)-Luc/35-mm-diameter plate. To investigate the effect of exogenous MTF-1 expression on basal PlGF transcription, the cells were cotransfected with pmPlGF(1.5kb)-Luc and varying amounts of a murine MTF-1 expression vector, CMV-MTF-1, which was created by inserting the mMTF-1 cDNA into the NotI site of a CMV expression vector (kindly provided by Dr. Glen Andrews; University of Kansas, Kansas City, MO). All transfections were normalized by cotransfection with a Renilla luciferase control vector (Dual-Luciferase Reporter Assay System; Promega). Where appropriate, a Bluescript phagemid (Stratagene) was used to ensure equal administration of total DNA. Cells were lysed, and luciferase activity was assayed by using a Promega Biotec assay kit and a Turner Designs TD-20/20 luminometer (Turner Designs, Inc., Sunnyvale, CA).
Basal Expression of PlGF in Mouse Fibroblasts.
During a cDNA microarray analysis of hypoxia-responsive gene expression in wild-type and MTF-1 null mEFs, we found that they express PlGF mRNA and that the PlGF gene is responsive to hypoxia in a MTF-1-dependent manner. Three PlGF forms are expressed in humans (PlGF-1, PlGF-2, and PlGF-3) (3), but only one isoform has been reported for the mouse (35). Therefore, we used conventional RT-PCR analysis to determine how many PlGF isoforms are expressed by mEFs. Total RNA was extracted from both aerobic and hypoxic wild-type mEFs and subjected to RT-PCR analysis, as described in “Materials and Methods.” The primers were selected to amplify sequences coding for a stretch of basic amino acid residues in exon 6 that confer the heparin binding ability of mPlGF (35) and hPlGF-2 (5) proteins. The hPlGF-1 and hPlGF-3 transcripts are produced by alternative splicing that removes the region coding for the basic amino acid residues responsible for heparin binding. In the case of hPlGF-1, this change shortens the coding region by 63 bases compared to that of hPlGF-2. The PlGF-3 transcript lacks this region but contains an additional 216-bp exon not found in the other forms (3). The RT-PCR primers were designed to detect all three of these known PlGF isoforms. Specifically, the primers were chosen to amplify a 147-bp fragment from the known mPlGF mRNA, a smaller 84-bp fragment corresponding to a homologue of hPlGF-1, and a 300-bp fragment corresponding to a homologue of hPlGF-3. This study showed that mouse fibroblasts express only one PlGF mRNA species, equivalent to hPlGF-2 (Fig. 1). The control lane (0; no RNA) in Fig. 1 shows no band, whereas the other lanes [Air and hypoxia (Hx)] contain a single band of the expected size (approximately 150 bp), corresponding to the size of the hPlGF-2 transcript. No bands corresponding to mouse homologues of hPlGF-1 or hPlGF-3 transcripts were detected.
Hypoxia Causes Increased PlGF mRNA Levels through a Mechanism Involving MTF-1 and Ras.
Fig. 2,A (top panel) shows a representative Northern blot of PlGF mRNA from H-ras-transformed wild-type and MTF-1 null mEFs that were subjected to hypoxia (pO2 ≤ 0.1%; 2–16 h). The figure clearly shows that mPlGF mRNA levels were significantly increased in the wild-type mEFs compared with the MTF-1 null mEFs by 4 h of hypoxia, with the greatest accumulation occurring at 16 h of stress. Fold inductions of PlGF mRNA accumulations could not be reliably determined because the aerobic steady-state levels of the mRNA were barely detectable. We also observed increases in PlGF mRNA levels over a wide range of low pO2 levels (≤0.1% to 2.0%) in the primary mEFs and in NIH 3T3 fibroblasts (data not shown). An analysis of two human cell lines (HepG2 and JAR) in which hypoxia was reported to have no effect on PlGF mRNA levels (e.g., see Ref. 10) was also included in these studies. As predicted from this earlier work, neither mild (pO2 ≈ 1–2%) nor severe (pO2 ≤ 0.1% O2) hypoxic exposures affected PlGF message accumulation in these experiments (data not shown). Northern analysis also showed that the induction of PlGF mRNA accumulation by hypoxia was strongly attenuated in the MTF-1 null mEFs (Fig. 2,A, top panel). However, there was a residual induction of PlGF mRNA accumulation in these cells, indicating other, less stimulatory levels of control of the PlGF gene in hypoxic mEFs. Finally, the accumulation of VEGF mRNA in both MTF-1 wild-type and knockout mEFs (Fig. 2 A, middle panel) was not significantly different, indicating that MTF-1 does not regulate the hypoxic response of this functionally related gene.
Fig. 2,B shows a representative Northern blot from a similar study using TAg-immortalized wild-type and MTF-1 mEFs. In this study, cells were subjected to two different levels of hypoxia (pO2 ≤ 0.1% and pO2 ≤ 0.01%) for 8- and 16-h exposures. Although significant inductions of PlGF mRNA levels were observed in the wild-type TAg mEFs, the detectable levels were much lower and sustained over a shorter period than those found in the corresponding wild-type mEFs transformed with a ras oncogene (see Fig. 2, A and B, top panels). This important finding suggests that hypoxia and oncogenic Ras can cooperate to induce PlGF expression mediated by MTF-1. A similar dependence involving oncogenic Ras was observed in the hypoxia-associated activation of the mouse metallothionein-I gene (data not shown). We have demonstrated elsewhere that two isoforms of MT are transcriptionally activated by hypoxia through a MTF-1-dependent mechanism (30).
Hypoxia Induces PlGF Protein Expression in a MTF-1-dependent Manner.
To determine whether the hypoxia-associated increases in PlGF mRNA correspond to PlGF protein inductions, conditioned media from H-ras-transformed wild-type and MTF-1 null mEFs subjected to 8 and 16 h of hypoxia (pO2 ≤ 0.1% O2) were analyzed for PlGF protein levels by standard immunoblotting procedures. Fig. 3 shows a representative immunoblot of PlGF protein secreted from these aerobic and hypoxic cultures. Secreted PlGF protein was undetectable in aerobic wild-type cells but was found in the hypoxic cell media with a strong induction by 16 h of hypoxia. In agreement with the Northern analyses, the relative induction of secreted PlGF protein by hypoxia was attenuated in the MTF-1 null mEFs but was not completely blocked. Again, a similar result was found for mouse MT-1 protein (30).
Isolation and Characterization of the mPlGF Promoter Region.
We successfully cloned and sequenced a genomic fragment of the 5′-flanking region of the mPlGF gene (GenBank accession number AF2855629) by using inverse PCR amplification, as described above. This fragment contains 1357 bp upstream of the largest known mPlGF cDNA sequence. The exact site(s) of transcriptional initiation is presently under investigation. A structural analysis of the promoter region was performed in an effort to detect putative MRE consensus sequences or other regulatory sites that may explain the gene’s responsiveness to hypoxia in wild-type but not MTF-1 null mEFs. Fig. 4 A shows the primary structure of the 5′-region of the gene (including additional sequences extending to the translation start codon) and indicates some possible transcription factor binding sites. Putative sites in the 5′-region flanking region of the gene include four sequences corresponding to a MTF-1 binding site (MRE) and at least five Sp1-like binding sites. Consensus sequences corresponding to AP-2 and tandem retinoid X receptor/peroxisome proliferator activated receptor-δ binding sites are also indicated in the figure.
A 203-kb region of human chromosome 14 has recently been sequenced that contains the gene for PlGF (GenBank accession number AC006530). The region containing the promoter of the hPlGF gene is shown in Fig. 4 B. A comparison of the human and mPlGF promoter regions shows both similarities and differences. For example, both promoters have two closely spaced Sp1-like sequences that are inverted relative to each other at about 400 bp upstream from the translation start codon. The hPlGF promoter also has four 9-bp repeats (CTGCGGGCT) within a 100-bp-region approximately 500 bp upstream from the translation initiation site. Two of these repeats are part of a 14-bp sequence (GGCGCTGCGGGCTG). It is not known whether these sequences are relevant for the transcriptional control of hPlGF, but they do have similarity to the core consensus sequence for the MRE, TGCRCNC (37, 38). This fragment of the human promoter contains five other possible MREs, of which two tandem copies (positions 500–521) display perfect MRE core consensus sequences in addition to G/C-rich 3′-flanking regions. Another putative MRE overlaps with a possible HRE at positions 279–290 (no consensus HRE was obvious in the mPlGF proximal promoter).
MTF-1 Is Involved in the Transcriptional Regulation of the PlGF Gene.
To define the molecular basis of the hypoxia-associated expression of the PlGF gene, we examined whether the increase in steady-state levels of PlGF mRNA in hypoxic mEFs is dependent on transcriptional activation. We cloned the mPlGF promoter fragment into a pGL3 reporter vector [pmPlGF(1.5 kb)-Luc; see “Materials and Methods”] and determined its responsiveness to hypoxia. The PlGF promoter reporter construct was transiently transfected into NIH 3T3 cells, and the cells were exposed to hypoxia (pO2 ≤ 0.01% for 16 h) before lysis to determine luciferase activity. Fig. 5 shows that this PlGF promoter reporter construct was significantly responsive to hypoxia (enhancements in luciferase activities relative to the aerobic controls were 4.5 ± 2.3; n = 4), indicating the presence of one or more HREs.
Finally, the involvement of MTF-1 in the regulation of PlGF expression was confirmed in studies using MTF-1 null mEFs that were cotransfected with pmPlGF(1.5 kb)-Luc and varying amounts of a mMTF-1 cDNA expression vector (see “Materials and Methods”). The basal transcriptional activities of the PlGF promoter reporter construct were significantly enhanced by ectopic expression of mMFT-1, with maximum increases of approximately 13-fold at 0.025 μg of pmMTF-1 (Fig. 6).
Hypoxia is established as an important environmental stimulus of angiogenesis in various pathological conditions, such as solid tumor development, some inflammatory states, and PDR. The induction of VEGF in response to hypoxia is an important mediator of angiogenesis in these pathologies (39, 40). The expression of other components of the proangiogenic response, such as VEGF receptors (flt-1 and KDR/flk-1) and angiopoietin-2, is also induced by hypoxia in various cells or tissues (7, 41, 42, 43). The present study extends these findings to show that the PlGF gene, a member of the VEGF family of proangiogenic factors, can also be activated by hypoxia. Specifically, we have demonstrated for the first time that immortalized/transformed fibroblasts (mEFs) express PlGF and that hypoxia causes increased expression of the PlGF gene in these cells by either a direct or indirect mechanism involving the transcription factor MTF-1. We have also shown that the induction of VEGF by hypoxia is not dependent on the presence of MTF-1 (Fig. 2). As discussed below, this induction of PlGF by hypoxia is potentiated by a ras oncogene.
Although PlGF expression has been considered to occur primarily in tissues composed of normal rather than transformed cells, several reports describe its expression in solid tumors, including hypervascular renal cell carcinomas (21), cervical squamous cell carcinomas (19), and some brain tumors (18, 20). PlGF expression has also been detected in a several established human tumor cell lines, including trophoblast choriocarcinoma cells (e.g., BeWo, JAR, and JE-3; see Refs. 10 and 27), HepG2 hepatoma cells (10), and U-25/MG glioma cells (20) and in melanoma cell cultures (44). Our study demonstrates that H-ras-transformed mEFs express at least one isoform of PlGF (PlGF-2). Studies of the effect of hypoxia on PlGF expression have yielded different results. For example, moderate hypoxia (pO2 ∼ 1–2%) was found to have no stimulatory effect on PlGF mRNA levels in a variety of cell types, including choriocarcinoma cells, HepG2 cells, dermal microvascular endothelial cells, retinal pericytes (9, 16, 27, 28), and cells in human placental tissue (28). In contrast, PlGF-1 and PlGF-2 mRNA levels were found to be induced in hypoxic U-25/MG brain meningioma cells (20). Induced PlGF protein expression has also been observed in ischemic retinal diseases [PDR, retinal vein occlusion, and acute retinal necrosis (22, 24)] and at sites of wound healing (11) and inflammation (17), both of which may involve hypoxic stress. Taken with our finding of strong hypoxia-inducible PlGF expression in immortalized/transformed mEFs, these other reports suggest that the response of the PlGF gene to hypoxia is not only cell- or tissue-type specific but is also susceptible to oncogenic stimuli.
Using mEFs with targeted deletions of both MTF-1 alleles, we observed that the hypoxia-inducible expression of PlGF is dependent on the presence of the MTF-1 transcription factor. Consistent with this finding, the PlGF promoter reporter studies described here indicate that hypoxia-inducible PlGF expression is controlled at least in part at the transcriptional level. The residual hypoxic response of the PlGF gene detected by both the Northern and immunoblotting analysis may be attributed to the contribution of other transcription factors and/or posttranscriptional regulation. Studies to determine the precise mechanism of MTF-1 involvement in the activation of PlGF by hypoxia are ongoing. In the context of oncogenesis, our findings suggest a role for activated ras in the transmission of hypoxic signals to MTF-1 with subsequent activation of PlGF expression. Interestingly, oncogenic ras can also cooperate with hypoxia to induce VEGF expression (45, 46), apparently through a mechanism that does not involve downstream activation of MTF-1 (see “Results”).
To our knowledge, this report is the first to show the sequence of the mPlGF proximal promoter region and the first sequence examination of the hPlGF promoter. The cloned fragment of the mPlGF promoter region contains 5′-sequences from the longest known mouse cDNA (GenBank accession number X96793), although the transcription start site remains undefined. An analysis of this longest cDNA fragment upstream from the translation start site shows that the GC content (57.8%) is only slightly enriched, and the CpG/GpC dinucleotide ratio (0.35) is well below what would be expected for a random sequence. These findings are compatible with the idea that the PlGF gene is not a housekeeping gene but is probably highly regulated because both high GC content and the lack of CpG dinucleotide suppression are considered common features of constitutively active promoters (47). A computer search for transcription factor binding sites identified four putative MRE consensus sequences, three of which are clustered within a 200-bp region from positions 1000 to approximately 1200 (Fig. 4,A). Two of these sites are within 15 bp of each other and just downstream from a putative Sp1 sequence. MREs are highly conserved 13–15-bp sequences that contain the heptanucleotide core consensus TGCRCNC and a partially overlapping, less conserved, GC-rich 3′-flanking region (37, 38). All four sites have almost perfect consensus core sequences with varying degrees of similarity in the flanking GC-rich regions. The hPlGF promoter contains at least five putative MREs (Fig. 4 B). Two perfect matches for the Sp1 core consensus site (GGGCGG) were found within 25 bp of each other in the proximal region of the mouse promoter. These two elements may be functionally important because they are also conserved in the human gene at approximately the same distance from the translation start site. Three other possible Sp1-like motifs (GGGC/T/AGG) are also present. Members of the Sp family of transcription factors are regulated by hypoxic stress though a mechanism involving relief of Sp3 repression of Sp1 (48). It is worth noting that both the human MT-IIA and mouse MT-I promoters also harbor Sp1 elements near their MREs (see Ref. 30). The possible regulatory roles of both the putative MRE and Sp1 sequences in the transcriptional activation of PlGF by hypoxia are currently under investigation. It is possible that MTF-1 activates PlGF by an indirect mechanism involving other downstream targets of MTF-1. For example, we have recently observed that MTF-1 regulates the expression of the transcription factor CCAAT enhancer-binding protein α.5
Sequence analysis also indicated that the cloned mPlGF proximal promoter fragment does not contain consensus sequences for the HIF-1 transcription factor (RCGTG). The hypoxia inducible factor-1 heterodimer is considered the major transcription factor affecting a wide range of hypoxia-responsive genes, including VEGF, erythopoietin, heme oxygenase-1, tyrosine hydroxylase, the urokinase receptor, nitric oxide synthase, and numerous glycolytic enzymes (e.g., see Ref. 49). However, other transcription factors, including AP-1, CCAAT enhancer-binding protein β, nuclear factor κB, and early growth response factor-1, have also been implicated in the cellular response to hypoxia (see Ref. 33). Taken with our previous work, the present study confirms that MTF-1 is another hypoxia-responsive transcription factor. Furthermore, we hypothesize that MTF-1 regulates a subset of hypoxic stress proteins that could be important for the adaptive response of transformed as well as normal cells to this stress. For example, we demonstrated that MTF-1 is required for the hypoxia-inducible expression of the human MT-IIA and mouse MT-I genes, critical determinants of Zn2+ homeostasis in mammalian cells (30). We postulated that MTF-1 is a redox-sensitive regulatory protein that influences malignant progression through control of MT expression and that of other known MTF-1 targets such as the hypoxia-responsive gene γ-glutamylcysteine synthase in the tumor microenvironment (30, 31). The findings presented here suggest that MTF-1 may also be an important regulator of tumor angiogenesis through its control of hypoxia-inducible PlGF expression.
In summary, we have shown that immortalized/transformed mEFs express PlGF and that exposure of these cells to either mild or severe hypoxia results in transcriptional activation of the PlGF gene. These findings imply that PlGF, a member of the VEGF family of proangiogenic factors, participates in the stimulation of tumor angiogenesis. Furthermore, we demonstrate that the hypoxia-inducible expression of PlGF is dependent on the presence of the MTF-1 transcription factor and is induced strongly in cooperation with oncogenic H-Ras.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by NCI Grants CA57692 (to B. J. M.) and CA73807 and CA67166 (to K. R. L.).
The abbreviations used are: PlGF, placental growth factor; hPlGF, human PlGF; mPlGF, mouse PlGF; HRE, hypoxia response element; MRE, metal response element; MT, metallothionein; MTF-1, metal response element-binding transcription factor 1; PDR, proliferative diabetic retinopathy; RT-PCR, reverse transcription-PCR; VEGF, vascular endothelial growth factor; mEF, mouse embryonic fibroblast; Tag, T antigen; CMV, cytomegalovirus.
P. Lichtlen, Y. Wang, T. Belser, O. Georgiev, U. Certa, and W. Schaffner. Target gene search for the metal-responsive transcription factor MTF-1, submitted for publication.